Message boards : Number crunching : Minirosetta 3.73-3.78
Author | Message |
---|---|
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
Please post any issues/bug reports regarding minirosetta and minirosetta_android 3.71 in this thread. |
Sid Celery Send message Joined: 11 Feb 08 Posts: 2112 Credit: 41,036,141 RAC: 20,543 |
New database coming down at 188Mb in size. One of my uploads struggling to upload. I anticipate everyone struggling to upload and download for a few days as the servers take a hit from everyone. David: what are the fixesimprovements in this version, please? Edit: 10mins to upload 2 tasks and download 4 3.71 tasks and the full new database took just 10 minutes here. Hopefully it won't be too bad for everyone, but still some patience required, obviously |
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
This version uses the latest Rosetta source code which includes an improved score function, new protocols (for example a new cyclic peptide modeling protocol), and some modifications to the graphics application. |
James W Send message Joined: 25 Nov 12 Posts: 130 Credit: 1,766,254 RAC: 0 |
One of my hosts is an Android 5 Tablet with 4 CPUs. Just a few moments ago I did an update and BOINC Event Log entry from Rosetta says: Rosetta Mini is not available for your type of computer. Is your server not yet recognizing the qualifications for 3.71? I know this message would be appropriate for earlier Rosetta Mini versions. Is the delay in sending 3.71 to my device due to overworked server or another reason? Thank you. |
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
We currently are not sending out the android work units yet. We will soon. |
Chilean Send message Joined: 16 Oct 05 Posts: 711 Credit: 26,694,507 RAC: 0 |
|
David E K Volunteer moderator Project administrator Project developer Project scientist Send message Joined: 1 Jul 05 Posts: 1018 Credit: 4,334,829 RAC: 0 |
Yes it is intentional. Someone in the lab wanted to make it look better but I guess it's all a matter of opinion. He also added a nice light source for a spacefill representation which you will see soon once he starts submitting his cyclic peptide modeling jobs. |
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
Yes it is intentional. Someone in the lab wanted to make it look better but I guess it's all a matter of opinion. He also added a nice light source for a spacefill representation which you will see soon once he starts submitting his cyclic peptide modeling jobs. Personally I like the new background coloring, though either way is fine really. Looks great in full screen. Will be interesting to see the cyclic peptide jobs. **38 cores crunching for R@H on behalf of cancercomputer.org - a non-profit supporting High Performance Computing in Cancer Research |
Professor Ray Send message Joined: 7 Dec 05 Posts: 35 Credit: 525,257 RAC: 181 |
This version uses the latest Rosetta source code which includes an improved score function, new protocols (for example a new cyclic peptide modeling protocol), and some modifications to the graphics application. This insinuated itself whilest crunching a v3.65 WU was retained in memory. Currently my machine is crunching WCG WU - both v3.65 & v3.71 Rosetta WU are RAM'd - prolly to swap space that's been striped across two U160 15k SCSI HDD. Its a pain in the neck - but I do it - when that v3.65 is done, I can gets rid of the old DB zip? And THEN only make symlinks to the newest DB? I've mentioned this before: I'd like better control over the DB extract process - so as to automate via script the symlink creation; but I was shouted down, belittled and trivialized by those superior to myself. I understand; that's the consequence of progress. :( That notwithstanding, once I can participate with 'progress' I'm certain I'll be a happy camper. ;) Imagine crunching on the video card, or even surpassing the 3.5GB memory constraint because of the hardware platform my O/S runs on! Oh, the future is such halcyion days! |
James W Send message Joined: 25 Nov 12 Posts: 130 Credit: 1,766,254 RAC: 0 |
We currently are not sending out the android work units yet. We will soon. Will we be notified, either by post or BOINC notice, when the WUs will be ready for sending? No use keeping space open on devices until work actually can begin flowing. Thanks. |
sgaboinc Send message Joined: 2 Apr 14 Posts: 282 Credit: 208,966 RAC: 0 |
updating, i tried this out on ralph actually, 3.71 in ralph seemed to run ok on linux. the blue background looks nicer but it seem somewhat stressful on the eye, i'd guess preference varies between individuals |
sgaboinc Send message Joined: 2 Apr 14 Posts: 282 Credit: 208,966 RAC: 0 |
no errors for 8 tasks |
Link Send message Joined: 4 May 07 Posts: 356 Credit: 382,349 RAC: 0 |
updating, i tried this out on ralph actually, 3.71 in ralph seemed to run ok on linux. the blue background looks nicer but it seem somewhat stressful on the eye, i'd guess preference varies between individuals Maybe they could make it configurabe thru Rosetta@Home Preferences, on SETI for example you can configure the screensaver that way. . |
sinspin Send message Joined: 30 Jan 06 Posts: 29 Credit: 6,574,585 RAC: 0 |
Krypton's works well since your changes and two days of runtime. Now, i got another WU with the same error ("exceeded elapsed time limit 141525.53 (500000.00G/3.53G)"): https://boinc.bakerlab.org/rosetta/result.php?resultid=788117823 I have about 1300 credits lost with this error, and two days of WU crunching of one core. |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
I have about 1300 credits lost with this error, and two days of WU crunching of one core. Credit for tasks that have errors is granted via a nightly script and, for whatever reason, is not shown on the task summary web page, only the task details. Rosetta Moderator: Mod.Sense |
Timo Send message Joined: 9 Jan 12 Posts: 185 Credit: 45,649,459 RAC: 0 |
Krypton's works well since your changes and two days of runtime. Target runtimes of 2 days have been deprecated for a while, to avoid issues I suggest moving your target runtime down to 1 day or less, this will provide enough buffer space between any hard limits like the one your WU hit. **38 cores crunching for R@H on behalf of cancercomputer.org - a non-profit supporting High Performance Computing in Cancer Research |
sinspin Send message Joined: 30 Jan 06 Posts: 29 Credit: 6,574,585 RAC: 0 |
Credit for tasks that have errors is granted via a nightly script and, for whatever reason, is not shown on the task summary web page, only the task details. Oh, good to know! Thanks. Target runtimes of 2 days have been deprecated for a while, to avoid issues I suggest moving your target runtime down to 1 day or less, this will provide enough buffer space between any hard limits like the one your WU hit. No, i will not move. I have a weak and unstable internet connection. I do not waste internet traffic with a shorter runtime and disk space as well. This bug is based on Boinc, not on the server bug that we had before. It seems that it need only a change in the parameters for a WU. I had before the same bug with *Krypton* WUs, since a change at server side there is no bug more. By the way, two days gives a high chance of getting 99 results per WU. And, wasting less crunching time for finalization of one WU and starting another. |
Link Send message Joined: 4 May 07 Posts: 356 Credit: 382,349 RAC: 0 |
Credit for tasks that have errors is granted via a nightly script and, for whatever reason, is not shown on the task summary web page, only the task details. Yes, but the maximum is 300 credits as you can see at the bottom of the result page. By the way, two days gives a high chance of getting 99 results per WU. And, wasting less crunching time for finalization of one WU and starting another. This is true, however the chance, that the WU errors out and all that work is lost increases too. . |
[VENETO] boboviz Send message Joined: 1 Dec 05 Posts: 1990 Credit: 9,482,262 RAC: 11,825 |
We currently are not sending out the android work units yet. We will soon. Will be some limitation, for example more recent than android 4.4?? |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
@sinspin, I would just point out that if files already exist on your host as you receive new work, then they do not have to be downloaded. However, if the host completes all tasks that need a file, the file is deleted. Any future tasks that might need it would have to download again. So, very difficult to predict your internet bandwidth requirement with much precision. Depends on the mix of tasks you get. But there are cases where having a larger pool of tasks on your host can improve the odds that you already have the needed files downloaded. If bandwidth and stability are chief concerns, you might consider installing something like squidproxy. It will stash a copy of downloaded files. Then if ever needed again, it already has a copy. Of course that would need some disk space to operate as well. But it (and longer runtimes) is the most effective way to save the network usage. Rosetta Moderator: Mod.Sense |
Message boards :
Number crunching :
Minirosetta 3.73-3.78
©2024 University of Washington
https://www.bakerlab.org